Adaptive-Size Block Transforms for Signal-Dependent Noise Removal

We present a new transform-based method for adaptive de-noising. It is assumed that the observations are given by a broad class of models with a signal-dependent variance. Denoising is performed by coefficient shrinkage in local block-transform domain. The intersection of confidence intervals (ICI) rule is used in order to determine the spatially-adaptive size of the block transforms. It enables both a simpler modeling of the noise in the transform domain and a sparser decomposition of the signal. Consequently, coefficient shrinkage is very effective and the reconstructed estimate´s quality is high. Experiments with simulated as well as with real data demonstrate the advanced performance of the proposed algorithm